By Derek Thomas, Vice President of Sales & Marketing, Machine Automation Solutions at Emerson
This article is the third in a four-part series exploring three new but related concepts for deriving value from big data: little data, edge processing, and embedded connectivity. The first column provided an overview of all three topics, and the second column focused on little data and how it differs from big data.
Many companies are tempted to tackle the big data issue in one fell swoop with mega-scale projects. A better path for most manufacturers is to start small with a little data approach to greatly simplify and speed implementation, with positive results generated in days instead of the years. Of course, one must keep in mind the ultimate goal, which is larger scale implementation, often through integration with enterprise IT systems.
A key enabling technology for creating value from little data is edge computing, whereby data produced by sensors is analyzed by a field-located device to generate insights. This information allows personnel close to the source to quickly assess issues and take appropriate action.
In the past, this type of edge processing would have required the addition of a separate industrial computing device, or an edge gateway connected to a server, to process the data. In either case, the solution would have required integration with the existing controller and manufacturing network. This would have been a problematic step due to the complexity of setting up and programming in two different environments—and due to synchronization requirements, lag/latency issues, cybersecurity concerns, and other factors.
But today, advances in processor technology enable edge controllers to perform two functions within a low energy use and compact form factor. The first function is real-time deterministic control using IEC 61131-3 languages, much like a traditional programmable automation controller (PAC). The second function is edge processing using advanced programming and scripting techniques. Integration effort, cost, and complexity are reduced because both functions are performed in one device.
Each part of the controller is virtually connected via OPC UA, allowing the two functions to be operate seamlessly, yet also separately such that real-time control tasks and performance are not affected by the edge applications.
PAC-based control is familiar to anyone working in the industrial automation arena, but edge processing contained in the same device is a new concept for many. In traditional implementations, a PAC would simply collect data and then forward it to a host system for processing. This host system would typically be PC-based, and it would often be located some distance from the edge, perhaps even in a data center or the cloud, raising a number of potential issues.
Processing data at the edge, instead of at a remote host, addresses these issues by providing local:
- Data storage, processing, and analytics
- Data logging
- Operational diagnostics
- Closed-loop optimization opportunities
- Visualization, dashboards, and other HMI functionality when connected to a display
Edge processing creates an open, flexible ecosystem with broad compatibility across all layers of automation systems. Users can create dashboards, improved analytics, and trends—all acting upon full fidelity and complete data sets—to empower local operators. Data communication and storage requirements are substantially reduced because only results are transmitted, instead of entire data sets.
With today’s PACs, customers can only write rudimentary analytics utilizing IEC 61131 languages. In contrast, edge controllers support execution of modern programing languages such as C/C++, Python, and Java. These and other advanced languages can be used to apply complex optimization algorithms or analytics to improve operations. This “outer loop” or “advise” layer runs in parallel with the underlying closed-loop control. In the event of a disruption to this “outer loop”, real-time deterministic control remains unaffected.
This functionality delivers on the potential of machine learning by allowing adjusts in real time based on actual performance data collected and analyzed at the edge. It also provides the ability to update machines/programs by staging upgrades, or sequencing them in real-time, to minimize impact.
An edge controller can thus be viewed as an end user’s on-ramp to the continuous journey of digital transformation. End users can start to solve big data challenges, one manageable little data chunk at a time, and edge controllers provide the scalability required to eventually transform entire discrete part manufacturing enterprises.
Sponsored content by Emerson